Mu opioid receptors on vGluT2‐expressing glutamatergic neurons modulate opioid reward

Abstract The role of Mu opioid receptor (MOR)‐mediated regulation of GABA transmission in opioid reward is well established. Much less is known about MOR‐mediated regulation of glutamate transmission in the brain and how this relates to drug reward. We previously found that MORs inhibit glutamate transmission at synapses that express the Type 2 vesicular glutamate transporter (vGluT2). We created a transgenic mouse that lacks MORs in vGluT2‐expressing neurons (MORflox‐vGluT2cre) to demonstrate that MORs on the vGluT2 neurons themselves mediate this synaptic inhibition. We then explored the role of MORs in vGluT2‐expressing neurons in opioid‐related behaviors. In tests of conditioned place preference, MORflox‐vGluT2cre mice did not acquire place preference for a low dose of the opioid, oxycodone, but displayed conditioned place aversion at a higher dose, whereas control mice displayed preference for both doses. In an oral consumption assessment, these mice consumed less oxycodone and had reduced preference for oxycodone compared with controls. MORflox‐vGluT2cre mice also failed to show oxycodone‐induced locomotor stimulation. These mice displayed baseline withdrawal‐like responses following the development of oxycodone dependence that were not seen in littermate controls. In addition, withdrawal‐like responses in these mice did not increase following treatment with the opioid antagonist, naloxone. However, other MOR‐mediated behaviors were unaffected, including oxycodone‐induced analgesia. These data reveal that MOR‐mediated regulation of glutamate transmission is a critical component of opioid reward.


| INTRODUCTION
The abuse of opioids, such as oxycodone, is on the rise in the United States, with more than 47,000 deaths due to opioid overdose in 2017. 1 Approximately one quarter of people prescribed opioids for pain management abuse them. 2 Chronic opioid abuse results in neurobehavioral changes including increased risk taking, impaired working memory, and impaired cognitive performance. 3 Mu opioid receptors (MORs) mediate the rewarding and analgesic effects of commonly prescribed and abused opioids. 4 It is well established that MORs modulate opioid reward through inhibition of GABA transmission. 5 The classical model of opioid reward implicates MOR-mediated depression of GABA release from VTA GABAergic interneurons that synapse onto local ventral tegmental area (VTA) dopamine neurons.
MOR activation decreases the inhibitory tone to VTA dopamine neurons, resulting in increased dopamine release in the nucleus accumbens (NAc). 5 This model is supported by the fact that ablating MOR expression in the VTA blocks the rewarding and stimulating effects of opioids. Others have suggested a revised model where MORs on GABAergic inputs from other brain regions to VTA dopamine neurons may mediate behavioral responses to opioids. In addition, other studies indicate that MORs expressed in forebrain GABA neurons also modulate opioid reward. 6,7 MORs also regulate glutamate transmission in multiple neurocircuits, including those involved in drug abuse, but their specific role in opioid reward is unknown. [8][9][10][11][12][13][14][15] It is increasingly clear that glutamate transmission is a critical component of opioid reward-related behaviors. The illicit opioid, heroin, increases extracellular glutamate in the NAc, an effect not seen with feeding-related behaviors. 16,17 Within the NAc and VTA, acute opioid exposure depresses glutamate release but enhances NMDA glutamate receptor function. Morphine fails to promote dopamine neuron activation following inhibition of VTA AMPA and NMDA receptors. 18,19 This treatment also inhibits the development of morphine conditioned place preference (CPP) without affecting locomotor activity. 20 We previously showed that MOR activation inhibits multiple glutamatergic inputs to the striatum, a critical brain region controlling behavioral responses to drugs of abuse. In these studies, we found that MOR activation inhibits glutamate transmission at Type 2 vesicular glutamate transporter (vGluT2)-expressing thalamic inputs to striatal medium spiny neurons. 21 vGluT2-expressing neurons are a subclass of glutamate neurons widely expressed throughout the brain. 22 vGluT2-expressing neurons in multiple brain regions modulate reward-related behaviors. [23][24][25][26] Given our findings that MORs regulate glutamate transmission at vGluT2-expressing synapses and the involvement of vGluT2 neurons in reward-related neurocircuitry, we decided to investigate the role of MORs in vGluT2 neurons in behavioral responses to opioids. Here, we report on the generation of a mutant mouse that selectively lacks MOR expression within vGluT2 neurons and its behavioral responses to the prescription opioid analgesic, oxycodone.

| Animals
Adult male and female C57BL/6J mice were purchased from Jackson Labs (JAX #000664). MORflox-vGluT2cre mice were bred and genotyped in-house from conditional MOR knockout mice (MORflox) and vGluT2cre progenitors (vGlut2Cre: JAX #016963). MORflox mice were generously donated by Dr Jennifer Whistler (UC Davis) and were previously characterized. 21 was computed and converted to a relative quantitative (RQ) value using the formula 2 −ΔΔCt .

| Electrophysiology
Brains slices were prepared for electrophysiological recordings as previously described. 21 Additional details may be found in the supporting information. Optically evoked excitatory postsynaptic currents (oEPSCs) in dorsal striatal medium spiny neurons were produced in brain slices using 470nm blue light (5 ms) delivered via field illumination through the microscope objective. Light intensity was adjusted to produce stable oEPSCs of 200 to 600pA amplitude prior to experimental recording. oEPSCs were evoked every 30 s. Prior to recording, brain slices were imaged via an Olympus MVX10 microscope Neurons were voltage clamped at −60 mV for the duration of the recordings. Data were acquired using Clampex 10.3 (Molecular Devices). Series resistance was monitored, and only cells with a stable series resistance (less than 25 MΩ and that did not change more than 15% during recording) were included for data analysis. Recordings were made 2 to 7 h after euthanasia.

| Behavior experiments
The following behavioral assays were performed in both male and female mice; studies were sufficiently powered to detect sex differences. For clarity of focusing on genotype, data presented in the main body of the text are collapsed across sex. Analyses of sex differences may also be found in supporting information. Detailed descriptions of behavioral assay methodologies may be found in supporting information.

| Oxycodone conditioned place preference
Mice underwent a modified protocol of oxycodone conditioned place preference (CPP). 28 Any mouse that showed an initial side preference greater than 200 s during the 20min pretest was given oxycodone in its initially nonpreferred side. Conditioning sessions (5 min) occurred twice a day for 3 days, with saline (10 ml/kg i.p.) conditioning sessions occurring in the morning (0900) and oxycodone (0.05, 0.5, or 5 mg/kg i.p.) conditioning sessions occurring 4 h later. A day after the final conditioning session, mice underwent a drug-free 20min test for CPP, comparing the amount of time spent in the oxycodone-paired and saline-paired sides.

| Open-field locomotor activity
Open-field chambers were used to measure baseline (10 ml/kg saline i.p.) and oxycodone-induced (5 mg/kg i.p.) locomotor activity. Each session was 20 min long.

| Naloxone-precipitated withdrawal
Oxycodone was administered using a modified dose ramping protocol (10 to 40 mg/kg over 8 days) previously demonstrated to produce oxycodone dependency in C57BL/6J mice. 29 Following development of oxycodone dependency, baseline (10ml/kg saline i.p.) and naloxone-precipitated (5mg/kg i.p.) opioid withdrawal-related behaviors were assessed for 10 min: paw shakes, wet dog shakes, jumps, ptosis, body tremor, teeth chattering, piloerection, and diarrhea. A global withdrawal score was calculated to give all withdrawal behaviors proportional weighting, as previously reported. 30 2.6.4 | Nociception/shock flinch Baseline (saline 10 ml/kg s.c.) and post-oxycodone (3 mg/kg s.c.) startle responses were measured in response to a variety of shock intensities.

| Two-bottle choice
Oxycodone, sucrose, and quinine consumption and preference were measured using a modified two-bottle choice (2BC) protocol with self-made ball-bearing sipper tubes, similar to those previously described. 31

| Food consumption
The amount of food in the home cage was weighed every 24 h, 11 h into the dark cycle, for 5 days. The amount of food consumed each day was calculated by subtracting the amount of food remaining from the previous day's food weight.

| Data analysis
Experimenters were blinded to genotype during all stages of data collection. Data are presented as the mean ± SEM. Data were analyzed using GraphPad Prism 8 (GraphPad, La Jolla, CA, USA). The level of significance was set at p < 0.05 for all analyses. Statistically significant individual data point outliers were identified using the ROUT method with Q = 1% and excluded. Some data were excluded on the basis of technical errors, such as leaking drinking tubes or equipment failures that occurred during a measurement session. Normal distribution was assessed prior to statistical analysis. Two-tailed unpaired t tests and two-tailed paired t tests were used to analyze normally distributed data. Mann-Whitney U tests were used to analyze non-normally distributed data. For data with multiple groups and/or repeated measures, ANOVA or restricted maximum likelihood (REML) with Sidak's post hoc tests was used. REML was used if data points were missing due to removal of outliers or experimental exclusion.

| MORflox-vGluT2cre mice lack MOR expression in vGluT2-expressing neurons
To ablate MOR expression specifically within vGluT2-expressing neurons, we bred conditional MOR knockout mice (MORflox) with mice that express Cre-recombinase in vGluT2-expressing neurons (vGluT2cre), similar to previous studies. 21,27 We assessed Creexpressing vGluT2 neuron MOR knockout mice and littermate controls, identified here as Cre+ (KO) and Cre− (Ctrl), respectively. The thalamus is a region with a high density of vGluT2-expressing neurons. 22  were also assessed in the cortex, a region with low vGluT2-expression. 22 In cortex, MOR expression was not significantly different between genotypes ( Figure 1B, Mann-Whitney test, p = 0.19). To functionally assess the knockout of MORs in vGluT2-expressing neurons in Cre+ (KO) mice, we investigated a vGluT2-expressing synapse we had previously shown to exhibit MORmediated inhibition of glutamate transmission. 21 Our previous work demonstrated that MORs inhibit glutamate transmission at vGluT2-expressing thalamostriatal synapses but did not reveal if it was specifically MORs within the vGluT2 neurons themselves that mediated that inhibition. 21 To address this question, as well as to functionally assess MOR knockout, we performed whole cell patch clamp electrophysiological recordings from dorsolateral striatal medium spiny neurons and specifically stimulated vGluT2-expressing thalamic inputs. To accomplish this, we injected AAV9.hSyn.ChR2

| MORflox-vGluT2cre mice lack oxycodoneinduced conditioned place preference and locomotor stimulation
MORs are involved in the locomotor and rewarding effects of opioids. 4 Low doses of opioids produce locomotor stimulation. 32 Therefore, we examined the locomotor activity of MORflox-vGluT2cre mice following an injection of either saline (10 ml/kg i.p.) or oxycodone  Table S1). Total fluid consumption did not significantly differ between genotypes ( Figure S3A). Females drank significantly more fluid overall than males, and there were significant differences in the amount of fluid consumed at each of the concentrations ( Figure S4C; Table S1).
Altogether, these data suggest that Cre+ (KO) mice do not find oxycodone as rewarding as Cre− (Ctrl) mice; however, an alternative hypothesis is that these mutant mice have altered preferences for rewarding substances or are possibly more sensitive to the bitter taste of oxycodone. Therefore, a separate group of mice underwent two series of 2BC: one with escalating concentrations of sucrose (0.5%, 1%, and 2% w/v), a naturally rewarding substance, and the other with escalating concentrations of quinine (0.03, 0.1, and 0.3 mM), a naturally aversive, bitter substance. In both series, the preference was tested against water and mice were counterbalanced in which order they underwent each series with a 1week washout period between each series. There were no genotype differences in sucrose consumption ( Figure 4C Table S1). There were also no genotype differences in quinine consumption ( Figure 4E Table S1). Cre+ (KO) and Cre− (Ctrl) mice also did not differ in food consumption, but there was a sex difference detected ( Figure S5A). Baseline weights were recorded for all mice involved in behavior experiments, before beginning testing. Cre+ (KO) mice weighed slightly less than Cre− (Ctrl) mice ( Figure S5B). Altogether, these results indicate that the genotype differences in oral oxycodone consumption are not due to general differences in sensitivity to naturally rewarding or aversive substances and further suggest that MORs on vGluT2Cre−expressing neurons are specifically involved with opioid reward.

| MORflox-vGluT2cre mice have intact oxycodone-induced antinociception
Because MORs are also involved in antinociception, 4 we measured responses to painful stimuli using the shock-flinch test. We used this   where sex differences were identified may be found in Figure S4 and Table S1.

| DISCUSSION
The major finding of this study is that MORs found on a subset of glutamatergic neurons that express the glutamate transporter, vGluT2, modulate opioid reward. In the absence of these vGluT2 neuron MORs, oxycodone reward is ablated. In addition, these vGluT2 neuron MORs mediate the locomotor stimulatory effect of opioid treatment, a behavioral response common to many drugs of abuse, 34 as well as opioidwithdrawal-related behaviors. Many studies find that MORmediated regulation of GABA transmission is critical for opioid reward, opioid-induced locomotion, and opioid self-administration. 5 The current data therefore suggest that there may be complex interplay between MOR-mediated regulation of GABA and glutamate transmission. Although MOR expression has been documented in glutamatergic neurons and MOR-mediated regulation of glutamate transmission has been reported for many brain regions, our findings demonstrate a role for MORs expressed in a specific subpopulation of glutamate neurons in modulating opioid reward. [8][9][10][11][12][13][14][15] Total MOR KO mice lack opioid-induced locomotor stimulation, opioid CPP, opioid withdrawal symptoms following naloxoneprecipitated withdrawal, and morphine-induced antinociception. 4  Cre− (Ctrl) mice. In total MOR KO mice, ethanol CPP is disrupted, whereas ethanol CPP is intact in Cre+ (KO) mice, suggesting that MORs in vGluT2 neurons do not mediate ethanol reward. 4,35 These data also suggest the behavioral effects of MOR genetic deletion from vGluT2-expressing neurons is specific to opioids, although additional drugs of abuse should be tested in the future. Food and sucrose consumption are also not affected in Cre+ (KO) mice and they weigh less than their Cre− (Ctrl) littermates, whereas total MOR KO mice have altered palatable food intake and show increased body weight. 6,[36][37][38] In forebrain GABAergic neuron-specific MOR KO mice, opioidinduced locomotor stimulation was ablated, similar to our Cre+ (KO) mice. 6 However, the GABA neuron MOR KO mice had increased opioid self-administration, intact opioid CPP, and reduced ethanol CPP, which all contrast with our results here. 6,35 A recent study also assessed opioid reward and aversion using mutant mice that lack MOR expression in the medial habenula. In these mice, behavioral responses to morphine were intact, but naloxone-induced withdrawal-related aversion was disrupted, further contrasting with our findings. 39 Altogether, our results suggest that MORs in vGluT2-expressing neurons have a critical role in modulating opioid reward.
One of the specific challenges moving forward is the identification of the specific vGluT2 neuron MORs that mediate opioid reward.
vGluT2 is expressed in primarily glutamatergic neurons and is predominantly expressed in the thalamus, amygdala, hypothalamus, cerebellum, and brainstem. 22 vGluT2 is also expressed in subpopulations of neurons in the VTA, ventral pallidum (VP), and certain regions of cortex. 22,40 Activation of vGluT2-expressing neurons within the VTA causes place preference, suggesting a role for vGluT2 VTA neurons in reward. 26 VTA vGluT2 neurons also play a role in aversion signaling, given that a subset of VTA vGluT2 neurons responds to aversive stimuli. 24 Studies have shown that vGluT2 VTA neurons drive aversion through local input to GABAergic interneurons in the VTA, as well as projections to the lateral habenula (LHb). 41 45 It is possible that MORs in these aversion processing vGluT2-expressing brain regions mediate opioid reward by inhibiting these aversion-encoding pathways. Glutamatergic projections from the rostral intralaminar thalamus to the dorsal striatum are involved in modulating reward through striatal dopamine release and we have both here and previously demonstrated that these synapses also express MORs. 21,46 It is possible that MORs regulate any of these sets of glutamatergic neurons to produce our observed behaviors and much work remains to be done.
It is important to note that while our intent was to study the behavioral role of MORs in glutamatergic neurons, our study is limited by the fact that vGluT2 is co-expressed in neurons that are known to release other neurotransmitters, that would not normally be considered classical "glutamatergic" neurons. For example, vGluT2 is expressed in VTA dopamine neurons, which co-release dopamine and glutamate; however, this represents a small population of vGluT2 neurons in the VTA. 47 In the brainstem, vGluT2 is expressed in catecholaminergic neurons. 48 vGluT2 is also expressed in cholinergic spinal cord motor neurons, which synapse onto muscles and cause them to contract. 49 In addition, vGluT2 can be found in GABAergic neurons in the anteroventral periventricular nucleus; a brain region mainly involved in sex-specific physiology and behaviors. 50 Given that most of these brain regions or neuronal subtypes are not part of what are usually considered reward neurocircuits, we do not think these are the likely sites of our behavioral effects. However, it is nonetheless possible that MORs in any one of these neural populations could mediate the effects seen in Cre+ (KO) mice. Future work is needed to determine which vGluT2-expressing neurocircuits are regulated by MORs and how MORs in these neurocircuits mediate opioid reward.
In conclusion, these MORflox-vGluT2cre mice are a valuable tool to dissect the role of MORs in a subset of glutamatergic neurons that express vGluT2. Our findings challenge the concept that MORs in GABA neurons are the principle drivers of opioid reward; although, they do not suggest that MORs in GABA neurons are not critical. We have demonstrated that there are vGluT2-containing neurocircuits that are involved in opioid reward and MORs expressed in these neurons mediate this reward. Although these mice have helped to reveal the role of these MORs in this class of glutamatergic neuron, additional work is needed to isolate the specific vGluT2 populations and brain regions responsible for these effects. Although this task will be challenging, it may ultimately lead to novel combinatorial pharmacological therapeutics for treating opioid abuse and addiction.